This aligns with what I've experienced. Been working on a large project for a big tech company (who is doing things in the AI space), and even when the content was speaking to Generative AI the creative director told us not to use any AI imagery because it looks cheap
I hope you're right but I feel for the vast majority of artists they will just have to change careers. There is only so much work an Apple could give out, so you'll have a few famous/established people getting work while the rest will have to compete with AI and eventually lose. As much as I would like to think people value quality, they value cost even more.
I think so too. I think for the vast majority if they want to stay in the industry they may have to transition to more of a pre-production role. Since production will be done by AI, someone will still need to come up with the script, storyboards, picking art-style, etc
Strongly suggest you wait till you get your hands on Sora before you lose hope. As someone who works with ML regularly I can tell you that these outcomes are great for pointed demos because the models are specifically trained on a small subset of subjects (which takes a vast amount of pre-processing and $$$). However, generalized, they don't do such a great job, especially with video because predicting subsequent frames for video is not something current ML models do very well. It's relatively easy to do predictive work with text, because there's a limited set of words. But with real world pixels, there's an explosion of possibilities and the predictive power of AI fails there. FYI these models don't understand physics - they are merely reproducing frame variations of frames they've already been trained on. It's still extremely impressive but not as ground-breaking as people are making it out to be.
Also.... a bit of a weird weird example: the McDonalds burgers... what a crap quality... and yes, it is one of the largest companies.... so true, AI will start to produce a lot of crap quality and there is a large audience for that. But the true burger shops, that create a good proper burger, are still around. And they will always be there. With customers that appreciate quality over 'speed and cost'
I see your point, AI in the hands of a bad director will still be bad. AI will be the equivalent of owning your own multi-million dollar VFX team, but as we see in mainstream cinema, that does not guarantee success.
It’s absolutely incorrect to compare handmade furniture & mass market --> and video generation. Because in the first case we have a material object in the real world, where you can notice the difference between expensive handmade materials and mass production. In the virtual world, however, we will soon not see this difference between AI and a human. Because they will be indistinguishable. There is no difference, whether the video is created by AI or a human, because the final format of this work will be - a video in a digital environment.
They may be indistinguishable, but HOW it was made matters a lot. For example, think about the value of a real Picasso vs an identical replica done by a skilled fraudster. One is priceless, and the other worthless. My thinking is that in the future, brands will make the extra effort into marketing how 'non-AI' their videos are. Kinda like how the new Top Gun movie & Oppenheimer heavily marketed how they did everything non-CGI (which eventually was not true) - perhaps that's a better analogy to have made in the video than furniture.
@@AlanPun2 You are comparing objects from the material world and the digital world again :) If you take an original work from NFT and then make a copy of it (Ctrl + C … Ctrl + V), apart from the registry entry that the first work is original, there will be no more value and difference. And I would speak carefully about the value of NFT. However, in any case, do not despair, these are all tools that will ultimately save us all time.
I enjoy your hopeful take on it and offering solutions to artists for working with AI, instead of the common ‘you’re all screwed’
Thanks Charlotte!
Great headline/image. I was like "damn, I need to watch this video!" then realized it was yours.
Thanks Kev!
"A.I. will not take over your job, the people that use A.I. will...."
haha true, true!
This aligns with what I've experienced. Been working on a large project for a big tech company (who is doing things in the AI space), and even when the content was speaking to Generative AI the creative director told us not to use any AI imagery because it looks cheap
hilarious
I believe that if you just continue in that pace you inevitably will go big with it, lots of sense of taste on each step and very bright mind.
nicest thing someone said to me all week! thank you!
And you quote Vonnegut! S5 is like my favorite novel ever :P
He's one of my favs too!
It's so over for us 3d world country folks
Its seems like its more over for them based on their current PR
I hope you're right but I feel for the vast majority of artists they will just have to change careers. There is only so much work an Apple could give out, so you'll have a few famous/established people getting work while the rest will have to compete with AI and eventually lose.
As much as I would like to think people value quality, they value cost even more.
I think so too. I think for the vast majority if they want to stay in the industry they may have to transition to more of a pre-production role. Since production will be done by AI, someone will still need to come up with the script, storyboards, picking art-style, etc
Strongly suggest you wait till you get your hands on Sora before you lose hope. As someone who works with ML regularly I can tell you that these outcomes are great for pointed demos because the models are specifically trained on a small subset of subjects (which takes a vast amount of pre-processing and $$$). However, generalized, they don't do such a great job, especially with video because predicting subsequent frames for video is not something current ML models do very well. It's relatively easy to do predictive work with text, because there's a limited set of words. But with real world pixels, there's an explosion of possibilities and the predictive power of AI fails there. FYI these models don't understand physics - they are merely reproducing frame variations of frames they've already been trained on. It's still extremely impressive but not as ground-breaking as people are making it out to be.
Thanks @@alankarmisra , do you anticipate that these downsides will eventually get ironed out due to the exponential growth of this field?
Also.... a bit of a weird weird example:
the McDonalds burgers... what a crap quality... and yes, it is one of the largest companies....
so true,
AI will start to produce a lot of crap quality and there is a large audience for that.
But the true burger shops, that create a good proper burger, are still around. And they will always be there. With customers that appreciate quality over 'speed and cost'
I see your point, AI in the hands of a bad director will still be bad. AI will be the equivalent of owning your own multi-million dollar VFX team, but as we see in mainstream cinema, that does not guarantee success.
It’s absolutely incorrect to compare handmade furniture & mass market --> and video generation. Because in the first case we have a material object in the real world, where you can notice the difference between expensive handmade materials and mass production.
In the virtual world, however, we will soon not see this difference between AI and a human. Because they will be indistinguishable. There is no difference, whether the video is created by AI or a human, because the final format of this work will be - a video in a digital environment.
They may be indistinguishable, but HOW it was made matters a lot. For example, think about the value of a real Picasso vs an identical replica done by a skilled fraudster. One is priceless, and the other worthless. My thinking is that in the future, brands will make the extra effort into marketing how 'non-AI' their videos are. Kinda like how the new Top Gun movie & Oppenheimer heavily marketed how they did everything non-CGI (which eventually was not true) - perhaps that's a better analogy to have made in the video than furniture.
@@AlanPun2 You are comparing objects from the material world and the digital world again :)
If you take an original work from NFT and then make a copy of it (Ctrl + C … Ctrl + V), apart from the registry entry that the first work is original, there will be no more value and difference. And I would speak carefully about the value of NFT.
However, in any case, do not despair, these are all tools that will ultimately save us all time.